DKS
Damn Kidney Stones
Premium,ExMod 2002
join:2001-03-22
Owen Sound, ON
clubs:
 ·Bell Sympatico
| reply to Bicephale
Re: A wire is a wire is a wire
said by Bicephale  :Now i know for sure that someone must have climbed up there much later than thirty-some years ago! I wonder if Bell would agree to insert an NID before the nuisances take place: e.g. on the roof, using only the wealthy section which remains once all of questionable items (like exposure to AC noise thru PVC tubing or the possible moisture-related shunt) have been excluded. Not likely. The NID is probably required to be accessible without a ladder.
--
Need-based health care not greed-based health care. |
|
ArthurS
Watch Those Blinking Lights
Premium
join:2000-10-28
Hamilton, ON
| reply to DKS
said by DKS :said by ArthurS :Looking back at earlier pictures, I don't like seeing the use of plastic tie wraps to secure the wire to the utility conduit, UV rays will deteriorate the plastic within a few years. Interesting. The Ilsco brand ties I have aree UV resistant. The ties were installed in direct sunlight five years ago on my satellite dish and they are still strong and flexible. Not saying that they aren't UV resistant tie wraps, but judging from the nature of the rest of the installation, all bets are off that such UV-resistant tie wraps have been used!  |
|
ArthurS
Watch Those Blinking Lights
Premium
join:2000-10-28
Hamilton, ON
| reply to Bicephale
The NID normally should be located where the wire enters into the building. You might find it a bit tough to convince the tech to replace everything, most have the attitude "that if it ain't broke, don't fix it"! Careful testing is the best way to show that there is a problem here, much of it can be attributed to sloppy workmanship inside and perhaps outside the house. Their responsibility ends at the NID, beyond that is on YOUR dime and responsibility! |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| reply to DKS
Bell's Main with Tape - Abrasion |
Hi ArthurS,
What's most interesting about the tie wraps is the
likelyhood that these were not part of the initial
installation when the house was built thirty years
ago. As i wrote before, this raises a fundamental
question relatively to the tape's purpose: was it
curative or simply preventive? In any case, there
is nothing left on our side of the demarcation box
which will require further attention! My expenses
reached a point where DSL isn't attractive anymore
if i consider that it made me buy two 25' rolls of
EtherNet cabling and two switches, one spare MoDem
because GNet bashers made me doubt without showing
the numbers, two RJ-11 phone connector boxes, iron
tubing (plus craftwork as a supplement), alligator
clips and an RJ-11 cord to match, one ground block
(to stick to the book), not to mention many months
of my own time spent over trouble-shooting... For
all i know, very few customers would even consider
going through all this extravaganza. Anyway, it's
impossible to set an NID over the entry point so i
suppose right above Hydro's counter should be good
enough - once the tape issue has been addressed...
By the way, maybe the tape was installed solely as
a protective ribbon but the image from above shows
that Bell may be better not to use tape like this!
 |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| reply to Bicephale
Ferro-magnetic Metal Sleeves | 
Bell's Main over Hydro's PVC Sections | | |
This is what's next on the menu: outdoor shielding, the last straw...
More numbers will be posted soon.
 |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
1 edit |  .PNG/thumb.jpg "20852 bytes")
Average Noise with Indoor Shield (Sep. 17 to Sep. 26, 2007) |
Here's what my indoor shielding accomplished... I
must confess i wasn't exactly expecting 2860 Local
CRC errors a day, especially with such sunny time!
At least my graph makes household activities (dish
washer, air conditionner, stove) even more easy to
recognize when i think of it; i also notice there
were few "unavailable seconds" compared to what it
used to be lately. Outdoor shield tests are next.
Addendum
Now that reminds me of something:
Feed-Back from a grateful customer!, Bicephale, 2007-Feb-15
Other than the UpStream speed tweak and the spikes
occuring late in the evening and at night (because
of heating, possibly), the peaks are comparable...
I wonder, could Bell's outdoor installation happen
to pick up so much 60 Hz noise through Hydro's PVC
tubing it doesn't really matter if i use shielding
inside or not (Bell's main line being saturated)?!
 |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| reply to Bicephale
%20%26%202nd%20Ground%20Bloc%20.JPG "161701 bytes")
Bell's Main (Shield Entrance) & 2nd Ground Bloc | %20%26%20Home%20Lines%20.JPG "116025 bytes")
Bell's Main (Shield Corner) & Home Lines | %20%26%20Home%20Lines%20.JPG "87797 bytes")
Bell's Main (Shield Exit) & Home Lines | 
Demarcation Point - Centralized Filtering |
Mild days are over, automn has arrived so i'll get
my last few photographs published. While i decide
when the next series of tests will take place lets
see what these modifications look like in the end!
 |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| reply to Bicephale
Wow! Wow Square! Wow to the nTh Power!
Take a good look at that post guys! Now i think i
may have found my guru and i'll worship the ground
on which he walks!... Quite refreshing! Amazing!
Speed slower than expected (Elite), d_l, 2007-Sep-16
 |
|
d_l
Barsoom
Premium,MVM
join:2002-12-08
Reno, NV
| reply to Bicephale
I guess I should explain a little about that graph. I have two DSL lines and have had CRC errors on both at times. I wanted to quantify how much various error rates affected the download throughput. I have several kitchen appliances, an ancient food blender, that I knew caused errors on both lines and another appliance, an electric oven/range that only affects the faster line.
I would run a large (16-64 MB) FTP download using an FTP program that times the download to the hundredth of a second. While the download was in progress, I then used the appliances to generate CRC errors. The modem's CRC counter was checked before and after the end of the FTP download. The throughput speeds of the error-impaired tests were normalized to the throughput of an error-free download.
CRC errors cause TCP/IP packet loss. So the download throughput speed during a burst of CRC errors is contingent on the recovery of the TCP/IP re-transmission which is in turn affected by the size of the TCP/IP window (I used a 63888 RWIN), the upload sync speed, etc.
Another factor affecting the download throughput speeds is the "clustering" of the CRC errors. A cluster of errors over a small fraction of a second may not cause as much slow down as the same number of errors evenly spaced over seconds time. I think my stove puts out more evenly spaced errors while the food processor's errors were more clustered. On the chart above, the stove was used to error the 6016 line and the food processor was only used on the 1536 line.
As they say, "your mileage may vary" from that in the graph. |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| Hi d_l,
I'm still without words, this demonstration put me
in a state of admiration for many reasons! I wish
my tests could compare but i realize that's beyond
reach for me, it's beauty expressed in a new form!
Well, i'll come to my senses eventually but i must
let you know that i appreciated my reading a great
deal. Right now it doesn't seem possible to get a
positive proof that i have a ground loop somewhere
but i'm thinking of using a battery to decouple my
MoDem from the AC circuit completely and then i'll
see if it changed anything. What is your opinion?
|
|
d_l
Barsoom
Premium,MVM
join:2002-12-08
Reno, NV
| reply to Bicephale
I've tested my DSL on back up battery power before and it didn't reduce my error rate when I intentionally tried to generate CRC errors. The impulse noise in my electrical lines from those noisy devices is transferred to the wiring in my phone drop by inductive coupling and not through the modem's power supply.
I would recommend that you try to find exactly which electrical devices are causing your errors if you can by powering up individual devices and checking the error counts in the modem. Once you know which devices (I'm assuming that not all devices will cause problems) are causing the errors, then you can jump to testing how much they might affect your download speeds using the method I explained above. When you know how much various error/rates affect your download speeds, then you can target a necessary reduction of the error rates that you can live with and/or afford to have what you consider acceptable download speeds.
You may find that the error rates you are currently seeing hardly affect your download speeds in which case you might be able to live with them. |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| Hi again d_l,
DSL devices are so sensitive i bet this technology
could be used to send messages to Mars, euh... It
wouldn't surprize me that parasitic capacitance is
enough to couple the MoDem to the 60 Hz AC circuit
through its EtherNet cabling, even if an isolation
transformer stands in the path. I wonder, does it
really suffice to run it on battery power or would
i not also need to unplug everything but the line?
This is wild speculation, all i can tell right now
is that the stove is most definitely involved. My
preliminary attempts to shield Bell's main outdoor
weren't conclusive - it's thinner, slotted and the
missing ground lead means that the metal sleeve is
"floating", after all, and requires a revision!...
Trying to measure the error rate simultaneously to
the transfer rate sounds like quite a challenge at
the moment. I'd need to make sure it's CRC errors
which cause lower speeds and i could identify only
one peak per hour so far. You see, i'm not facing
one but two issues; sometimes the phone line will
keep steadily quiet for days with the exception of
kitchen activity but there are days when the noise
is most likely external to our home... During bad
days it doesn't help to measure noise because this
disturbance coming from nowhere will mask reality,
no matter if the stove is being used or not. I've
considered buying a new MoDem, the 4th one i'd try
so far, but my experience with the SpeedTouch left
me with the opinion that newer products just carry
economically-based design optimisations if any!...
You mentioned the SpeedStream 4100B because it has
better rejection of the noise induced electrically
by household appliances. Doesn't this device sell
as a "decomissioned" (customized) refurbished unit
from Bell, etc., or are there suppliers who happen
to have some SS4100B with generic FirmWare? 'DMT'
is a precious tool for me, would it work on it?...
|
|
Angelo_
The Network Guy
Premium
join:2002-06-18 | reply to Bicephale
map the wire :P |
|
Angelo_
The Network Guy
Premium
join:2002-06-18 | reply to Bicephale
what do you got against the 516 its so great much better then my 3com i bought... and let me tell you what a waste taht was! |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
2 edits | Hi Angelo,
Do you mean you'd like me to post a drawing of the
phone line configuration? It seems straigtforward
compared to others, there isn't much more to tell;
well, there's a couple inches of wiring going from
Bell's old demarcation box to that white connector
box on its left and i've already explained how the
filters & MoDem connect, euh... but here it goes:
%20.PNG)
This isn't how it looks physically but that's just
equivalent electrically speaking if we except that
ground link which i forgot to implement this time.
About the SpeedTouch, you may have noticed that my
appreciation of it differs from the mainstream and
i can't say otherwise: put shortly, it's inferior
to my GNet in every way and i've shared numbers in
previous posts which progressively proved that its
reputation sounds a bit over-exagerated sometimes.
The very first link at the begining of the present
thread provides various tips relatively to, euh...
lets call it my discomfort about Thomson's ST5x6v6
series. See how a GNet will raise my UpStream SNR
Margin from 8 dB to 22 dB - imagine! - while there
are many such features forbiden to the SpeedTouch.
As i wrote before, i simply can't agree with those
who would qualify it with terms like "trash", etc.
but i don't share the opinion of others saying the
SpeedTouch is a winner neither. I think it's time
to update our collective appreciation of what's on
the market months after the frenzy has past. I do
appreciate your comment about the 3Com, is it from
the USR 910x series? I must confess that i happen
to be curious about the TP-Link units too and i've
asked for reviews on it (well, at least twice, now
that i think of it)... In any case, i welcome all
of the information that people can post on MoDems!
If it's still permissible on Canadian BroadBand...
Addendum
Maybe more clues are required, this should help a bit:
ST546v6, 2007-Mar-12: 4899 errors/day (calculated over 2½ days)
ST546v6, 2007-Jun-6: through the roof!
|
|
TSI Steve
TSI Steve
Premium,VIP
join:2007-01-12
Chatham, ON
| said by Bicephale :About the SpeedTouch, you may have noticed that my appreciation of it differs from the mainstream and i can't say otherwise: put shortly, it's inferior to my GNet in every way and i've shared numbers in previous posts which progressively proved that its reputation sounds a bit over-exagerated sometimes. Hello Bicephale,
We've been noticing your posts about the differences between the SpeedTouch and GNet, to that end we just wanted to let you know that in some cases we agree that the GNet out preforms the SpeedTouch.
For example:
In *your* specific case, when the distance is less then 3/3.5KM the GNet does on occasion out preform the SpeedTouch DSL modems. However, if you get the link up over 3/3.5KM mark the GNets reliability degrades very rapidly.
Overall, for the price GNet's are a decent product when you are less then 3/3.5KM.
I have personally dealt with a few thousands clients that specifically bought GNets. In almost all of the cases where the clients were less then 3.5KM the GNet worked just fine. In the cases where the distance is over 3.5KM we see a lot of issues arise with stability of the link, errors occurring, and the device overall not functioning correctly.
Don't get me wrong, I am not here to bash GNet. We don't bash other companies products. I will note the fact that when we were selling GNet's the return rate for defective devices was around 30% (or more). The return rate on the SpeedTouch 516 is around 2%. The SpeedTouch 546 it's less then 1%. The SpeedTouch 780 (typically related to it's VoIP support) is around 15%. Those numbers are much more manageable.
--
TSI Steve - TekSavvy Solutions Inc. |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| Hi TSI Steve,
Thanks a lot for sharing your feedback! There's a
question which i wanted to ask to people with your
kind of expertise for some time now: can you tell
in approximate terms what proportion of the return
rate was due to flash accidents? Another one that
i've been thinkering about lately is what would it
take for Thomson and TekSavvy to make an agreement
over the development of custom FirmWare which will
make the SpeedTouch 5x6 equivalent or better, even
when distance is less than 3½ Km as in my case?...
 |
|
d_l
Barsoom
Premium,MVM
join:2002-12-08
Reno, NV
| reply to Bicephale
The 4100b is probably the equivalent of a generic 4100 modem with a recent firmware. The big difference of the AT&T-versioned 4100b vs. their 4100 was the change to support ADSL2+ in the firmware. Generic 4100 modems are somewhat pricey compared to AT&T versioned 4100 modems on eBay. Because AT&T modems are so common, the prices tend to be lower. If you would use the eBay route to purchase a 4100b, be very careful that it really is a 4100b version.
A straight up purchase of the modem directly from AT&T would cost you $49 plus S&H: »store.att.com/Catalog/ProductDet···d=catMRG
Although your DMT software wouldn't work on the 4100b because the modem's CLI interface isn't accessible when the modem is in DSL use, we have devised a tool to plot the tone bins from the html interface: »SBC DSL FAQ »How can I check for the maximum attainable sync speeds with a 5100b/4100 modem?. Also we have other diagnostics to determine the presence of bridged taps: »SBC DSL FAQ »What do the hybrid numbers mean and what do they tell you about your line?. |
|
TSI Steve
TSI Steve
Premium,VIP
join:2007-01-12
Chatham, ON
| reply to Bicephale
said by Bicephale :Hi TSI Steve, Thanks a lot for sharing your feedback! There's a question which i wanted to ask to people with your kind of expertise for some time now: can you tell in approximate terms what proportion of the return rate was due to flash accidents? Another one that i've been thinkering about lately is what would it take for Thomson and TekSavvy to make an agreement over the development of custom FirmWare which will make the SpeedTouch 5x6 equivalent or better, even when distance is less than 3½ Km as in my case?... In the case of the GNets, flashing is almost always successful. I literally only dealt with a handful that were due to flashing issues (I am excluding the cases where I was able to manually TFTP in using the serial connection to upload new f/w). On the SpeedTouch side (780 more then any) we've seen a higher error ratio on flashing 'accidents'.
We've attempted to work with SpeedTouch on a couple custom things in the past (in regards to the 780 and VoIP) they responded very well to our needs. It may just be a matter of coming with a good plan and presenting it to them and they may implement it.
Steve
--
TSI Steve - TekSavvy Solutions Inc. |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
1 edit | Hi TSI Steve,
I ask because i've had to return my 1st GNet after
a DownGrade... What i like about the GNet BB0060B
is that i can generate 24 h Error Rate curves with
the help of 'DMT' once a day. 'STMT' was supposed
to be able to display a 24 h Error Rate curve also
but it must be left running continuously and i can
not make it work for me, not to mention i wouldn't
like to leave a machine running for one whole week
in order to get one single 7 days record... If at
least the SpeedTouch could send its raw data every
15 minutes thru e-Mail so that external processing
generates curves, euh... it would be good enough.
By the way, a Linux Live CD using SNMP support may
suffice but i don't see that anywhere i look. The
guys at Thomson should explore this avenue, maybe!
Another thing i like relatively to my GNet is that
each bandwidth is adjustable: it allows me to set
the number of tones and their weight - that's what
i refered to as being called "Spectral Shaping" in
previous posts... Finally, i also wrote about the
Twin PPP configuration and i even got some replies
from TekSavvy but that part didn't work. I had to
give up on it but i'm open to future improvements:
when a LogIn account is available it helps to read
the log if two PPP sessions are established to two
different ISPs; it's how i came to know tekSavvy,
as i recall (Rocky offered me a temporay LogIn)...
Can i hope for a custom FirmWare at ChristMass?...
Addendum
It seems my thread was interrupted by the system but it continues here:
A wire is a wire is a wire, Bicephale, 2007-Oct-13
|
|
